Cornell Research Foundation Inc.

Cornell University

José F Martínez

Helena Caminal

Christopher Batten

Meyrem Kirman

Kailin Yang

Nevin Kirman

Khalid Al-Hawaj

Engin Ipek

A content-addressable processing engine, also referred to herein as CAPE, is provided. Processing-in-memory (PIM) architectures attempt to overcome the von Neumann bottleneck by combining computation and storage logic into a single component. CAPE provides a general-purpose PIM microarchitecture that provides acceleration of vector operations while being programmable with standard reduced instruction set computing (RISC) instructions, such as RISC-V instructions with standard vector extensions. CAPE can be implemented as a standalone core that specializes in associative computing, and that can be integrated in a tiled multicore chip alongside other types of compute engines. Certain embodiments of CAPE achieve average speedups of 14× (up to 254×) over an area-equivalent out-of-order processor core tile with three levels of caches across a diverse set of representative applications.

Content-addressable processing engine

A content-addressable processing engine, also referred to herein as CAPE, is provided. Processing-in-memory (PIM) architectures attempt to overcome the von Neumann bottleneck by combining computation and storage logic into a single component. CAPE provides a general-purpose PIM microarchitecture that provides acceleration of vector operations while being programmable with standard reduced instruction set computing (RISC) instructions, such as RISC-V instructions with standard vector extensions. CAPE can be implemented as a standalone core that specializes in associative computing, and that can be integrated in a tiled multicore chip alongside other types of compute engines. Certain embodiments of CAPE achieve average speedups of 14&#xd7; (up to 254&#xd7;) over an area-equivalent out-of-order processor core tile with three levels of caches across a diverse set of representative applications.

A reconfigurable multiprocessor system including a number of processing units and components enabling executing sequential code collectively at processing units and enabling changing the architectural configuration of the processing units.

Systems and methods for reconfiguring on-chip multiprocessors

A technique known as checkpointed early load retirement, combines register checkpointing load-value prediction to manage long-latency loads. When a long-latency load reaches the retirement stage unresolved, the processor enters Clear mode by (1) taking a Checkpoint of the architectural registers, (2) supplying a load-value prediction to consumers, and (3) early-retiring the long-latency load. This unclogs retirement, thereby 'clearing the way' for subsequent instructions to retire, and also allowing instructions dependent on the long-latency load to execute sooner. When the actual value returns from memory, it is compared against the prediction. A misprediction causes the processor to roll back to the checkpoint, discarding all subsequent computation.

Growing community of inventors

Ithaca, NY, United States of America

José F Martínez